Neural Network Control
Modern cyber-physical systems have complex dynamics that are hard to model and optimize for. The ability to integrate low-power, low-weight high-performance embedded microprocessors enables data-driven model estimation and control refinement. In our lab, we are pushing the boundaries of NN-driven flight control with an emphasis on system deployability, generalizability, and safety assessment.
Projects
7 publications in this area
- Accelerating Lyapunov-Stable Neural Control using Fulfillment Priority Logic
- Closing the Intent-to-Reality Gap via Fulfillment Priority Logic
- Unified Local-Cloud Decision-Making via Reinforcement Learning
- Honey. I Shrunk The Actor: A Case Study on Preserving Performance with Smaller Actors in Actor-Critic RL
- Regularizing Action Policies for Smooth Control with Reinforcement Learning
- How to Train your Quadrotor: A Framework for Consistently Smooth and Responsive Flight Control via Reinforcement Learning
- Reinforcement Learning for UAV Attitude Control